Why tracked-user pricing punishes you for succeeding

We sell a feedback tool, so this is not a neutral essay. It is still arithmetic you can check, and the arithmetic is the point.

The arithmetic

Take Canny, the clearest example because its figures are public. Free covers 25 tracked users. Core starts at $19 a month, Pro at $79 billed annually. A tracked user is anyone who submits, votes or comments.

Now run a good quarter. You launch something, traffic doubles, and the share of customers who bother to vote on a request goes from tiny to merely small. Reported figures at 1,000 tracked users are around $275 a month on Core and $579 on Pro. Nothing about your revenue necessarily changed. What changed is that more people participated.

The uncomfortable part is the incentive it creates. Every prompt that gets more customers to vote makes your tool more expensive. Teams respond rationally: they stop promoting the board.

Per-seat has the same shape, aimed at your team

Featurebase charges $29, $59 and $99 per seat per month on its paid tiers, plus $0.29 per AI resolution. That is predictable, which is a real advantage over engagement-based metering. Its failure mode is pointed at the other side of the table.

Customer feedback is most useful when the people building the thing read it directly. Per-seat pricing puts a price on exactly that, and the result is familiar: one person has the login, exports a summary, and pastes it into Slack. The information gets one layer of lossy compression added because the compression is cheaper than a seat.

Per-session bills you for traffic

Behavioural tools like Hotjar meter on sessions, from around $32 a month on paid plans. If your question is genuinely behavioural that is reasonable, since sessions are the raw material. It does mean a piece of content going unexpectedly viral shows up as a bill.

What to look for instead

• Can you forecast it? You should be able to estimate next quarter's bill from numbers you already have. Tracked users fails this: nobody knows what share of their users will vote.
• Does it punish the behaviour you want? If getting more feedback, or getting more of your team to read it, costs extra, you will do less of both.
• What happens at 10x? Multiply the meter by ten and look at the number. If it is absurd, you will be migrating during the busiest quarter you have ever had.

For what it is worth, Voicebox meters on feedback collected: free for 50 a month, $19 for 3,000, $49 for 15,000 with unlimited seats. That has its own failure mode, which is a spam wave, and the fix is the domain allowlist rather than a bigger plan. Every meter fails somewhere; the useful question is whether it fails when things are going well or when things are going badly.

If you are weighing specific tools, the Canny alternatives piece works through the same numbers by product.

Common questions